fftw_flags — set computation method of fast fourier transform of the fftw function
[a,[S]]=fftw_flags([x1;x2;...])
Matrix of string or integers. Entry to switch the method of fft computation for fftw.
Integer. Give the current value of the flag of the fftw function.
String matrix. Give the string value of the fftw flag.
This function enables the change of the unsigned
flags
parameter of the
fftw_plan_guru_split_dft
function that is used in fftw function.
Default value is FFTW_ESTIMATE
Accepted entries are :
FFTW_MEASURE or 0
FFTW_DESTROY_INPUT or 1
FFTW_UNALIGNED or 2
FFTW_CONSERVE_MEMORY or 4
FFTW_EXHAUSTIVE or 8
FFTW_PRESERVE_INPUT or 16
FFTW_PATIENT or 32
FFTW_ESTIMATE or 64
FFTW_ESTIMATE_PATIENT or 128
FFTW_BELIEVE_PCOST or 256
FFTW_NO_DFT_R2HC or 512
FFTW_NO_NONTHREADED or 1024
FFTW_NO_BUFFERING or 2048
FFTW_NO_INDIRECT_OP or 4096
FFTW_ALLOW_LARGE_GENERIC or 8192
FFTW_NO_RANK_SPLITS or 16384
FFTW_NO_VRANK_SPLITS or 32768
FFTW_NO_VRECURSE or 65536
FFTW_NO_SIMD or 131072
FFTW_NO_SLOW or 262144
FFTW_NO_FIXED_RADIX_LARGE_N or 524288
FFTW_ALLOW_PRUNING or 1048576
Rmk : when using FFTW_MEASURE/FFTW_PATIENT/FFTW_EXHAUSTIVE you must call two times fftw. (first call for initialisation, second and others calls for computation)